For diseases without safe and long-term effective therapies, such as obesity, human genetics offers an unbiased route to biological insights that may provide valuable new therapeutic hypotheses. Genome-wide association studies (GWAS) have implicated both known and novel genes for many polygenic traits, including obesity. However, moving from genetic discovery to biological insight requires overcoming some key hurdles. Because associations from GWAS typically identify clusters of correlated noncoding variants, associated loci typically do not pinpoint either specific regulatory elements or causal genes. In addition, little is known about the function of most genes, so it is often difficult to recognize the biological implications of new discoveries. Fortunately, there is a path forward ? considering associated loci in combination can reveal shared biology and causal genes not apparent from any individual association ? but powerful computational methods and large numbers of associated loci are needed for this approach to work. For height, a model polygenic trait with many known loci, this approach highlights many relevant pathways and genes, both known and novel. Similar insights have only just begun to emerge when applied to measures of obesity, where there are fewer known loci and likely less well-annotated causal biology. The main goal of these genetic studies is to achieve a clearer view of underlying biology, and progress has been more dramatic for height than for obesity. As such, the current success with height shows the promise for a greatly expanded genetic discovery effort for obesity. This proposal aims to fulfill the promise of human genetics to provide critical insights into the root biological causes of obesity. It builds on the collaborative infrastructure we successfully created within the GIANT consortium and have used to discover most of the common variants known to be associated with anthropometric traits. The work will leverage newly feasible genetic approaches and unprecedented sample sizes to study anthropometric measures of obesity (a major public health problem and unmet medical need) and height (the classical model polygenic trait). To increase the number of genetic discoveries, which is vital to recognizing underlying biology, the proposal encompasses the largest collection of genotyped samples yet assembled (up to 2 million individuals from multiple ancestries), imputed to state-of-the-art reference panels. Association analysis for anthropometric traits will also be performed in large whole genome and whole exome sequence data sets (N>100,000), to discover rare variants that may have larger effects and more precisely pinpoint causal genes/regulatory elements. Computational methods that integrate genetic, expression and epigenetic data will be benchmarked on results from height, and then applied to recognize shared biology across obesity-associated loci and across the allelic spectrum, providing insights into likely causal genes and mechanisms. Finally, Mendelian randomization will be used to infer causal relationships between obesity and circulating metabolites, to define metabolic consequences of obesity as well as new therapeutic opportunities.